A model for calculation of steam injector performance

N. Deberne
a,
*, J.F. Leone
a
, A. Duque
b
, A. Lallemand
a
a
CETHIL, UPRESA CNRS 5008, INSA Lyon, 20 avenue Albert Einstein, 69621 Villeurbanne cedex, France
b
EDF/SEPTEN, 1214 Avenue Dutrievoz, 69628 Villeurbanne Cedex, France
Received 4 August 1997; received in revised form 20 October 1998
Abstract
Steam injectors can be used in many applications, but especially for security water injection in steam
generators of nuclear reactors. Using a one-dimensional model, a steam injector with a centered liquid
supply has been simulated. General relationships are presented from the nozzle exit to the steam injector
outlet. It is shown that the ow contains a condensation shock. To achieve modelling of the mixing
zone, and empirical correlation giving an equivalent pressure with value of condensation rate is found
using experimental results obtained at the CETHIL. A parametric study is then made to determine the
inuence of signicant parameters and the functioning range of the steam injector. Calculated values are
compared with experimental results and are found to be in good agreement. # 1999 Elsevier
Science Ltd. All rights reserved.
Keywords: Steam injector; Jet pump; Condensing injector; Ejector condenser
1. Introduction
A steam injector is a device whereby steam is used to pump cold water at low pressure and
to produce an outlet water pressure which is higher than the steam inlet pressure. Its main
characteristic is that no moving parts are needed for its functioning, thermodynamic processes
relying on direct transfers of mass, momentum and heat between the two phases.
The steam injector can be used as a safety pump in a light water reactor, as a steam supply
is generally available in power plants and a high pressure water supply can be useful for heat
removal in case of incident. Moreover, it is a passive system without rotating machinery and
its functioning requires no external energy supply.
International Journal of Multiphase Flow 25 (1999) 841855
0301-9322/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0301- 9322( 98) 00071- 8
PERGAMON
www.elsevier.com/locate/ijmulflow
* Corresponding author.
A simplied steam injector schematic, divided into ve regions, is shown in Fig. 1. In the
rst part, called the steam nozzle (a), having a converging-diverging shape, the steam is
accelerated to supersonic velocity through a nearly isentropic expansion. Water is distributed
by the water nozzle or the liquid injector (b). Here, the chosen arrangement is a central water
nozzle and an annular outer steam nozzle. In our study, it will be seen that the geometrical
arrangement has an eect on the results, but it can be inverted with a small modication to the
model. In the mixing section (c), steam and water exchange heat, momentum (due to
temperature and velocity dierences) and mass (due to condensation of steam on the water
droplets extracted from the water cone at the exit of the water nozzle). Condensation is
achieved in the shock wave (d) occurring at the exit of the mixing section. The major pressure
rise is realized in this shock wave. Then, the water is decelerated in the diuser (e) and kinetic
energy is converted into a further pressure increase.
Although the technology of the steam injector has been known for about a century, its
modelling still represents an incompletely solved problem. Experiments have been previously
carried out by Rose (1960), Grolmes (1968) and Alad'yev et al. (1981a, b). More recently, its
possible use for nuclear power plants has involved new studies by Manno and Dehbi (1990),
Narabayashi et al. (1991), Leone et al. (1994, 1995) and Cattadori et al. (1991).
Two kinds of model can be used for the calculation of steam injector performances. The rst
approach (called the local model) consists of modelling the ow in the injector at each point,
taking into account major phenomena (such as, for example, condensation of droplets, heat
and momentum transfers, viscous dissipation on the wall non-adiabatic and non-equilibrium
ow). Such a model is more realistic, but needs a perfect knowledge of the previously
mentioned phenomena as well as their mutual interactions. Unfortunately, these models are
not yet predictive due to the complexity of the ow in the mixing chamber.
The second approach (called the global model) consists of considering a large control
volume in which one-dimensional conservation equations are applied. This kind of model is
simpler than the previous one but is less realistic. Irreversibilities are often taken into account
using empirical factors (in order to obtain calculated data tting experimental values), which
cannot really be justied. Moreover, these factors are only known for one experimental
condition and cannot be generalized.
In this paper, a global model is employed but, instead of using correcting empirical factors,
irreversibilities are taken into account through the pressure variation along the mixing section.
It appears that, for the central water nozzle arrangement, pressure variations can be easily
modelled depending on the condensation rate in the mixing section. Moreover, the two phases
Fig. 1. Schematic sketch of steam injector.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 842
are considered separately and the shock wave and the mixing section are modelled apart.
Closure of the equation set is ensured by equilibrium assumptions at the exit of the mixing
section and by a condition of complete condensation downstream of the shock. Properties of
water are calculated with software created at the CETHIL based on the Schmidt (1969)
formulation. This approach is entirely original and can be used without acquiring any previous
experimental data.
A parametric study is then used to determine the inuence of signicant parameters and the
functioning range of the steam injector. Calculated values are compared with experiments
performed at the CETHIL and are found to be in good agreement.
2. Steam injector modelling
Modelling of the steam nozzle (a) and water nozzle (b) is not presented in this paper, as the
uids states depend on the arrangement of pipes in the rig. In the convergingdiverging steam
nozzle, a supersonic ow is obtained and steam state is supposed to be known at the mixing
section inlet. The subscript referring to this plane will be noted 1s. In the same way, subscripts
1l, 2, 3 and 4 will respectively dene the water mixing section inlet (water nozzle outlet), the
mixing section outlet, the shock wave outlet and the diuser outlet planes. It should be pointed
out that in plane (2), both steam and water are present so subscripts 2s and 2l will be used, but
will not correspond to separate phases as in (1s) and (1l). The ow is assumed to be steady and
one-dimensional.
2.1. Modelling of the mixing section
The mixing section is the most important part of the steam injector, where momentum, mass
and heat transfers between phases take place. Two kinds of ows can be observed in the
mixing section: a stratied ow in the upstream region, consisting of a liquid jet surrounded by
wet steam and a downstream homogeneous two-phase ow. The liquid jet is gradually
disintegrated into small droplets by the supersonic ow of steam until it completely vanishes.
In this rst region, momentum transfer is very important. When the ow is a homogeneous
mixture of liquid and steam, the exchange surface is greater and the mass and heat transfers
are thus increased.
In our system, pressure is assumed to be equal to the steam pressure, which is a well-known
assumption in most two-phase ow congurations. Consequently, it is assumed P
S
=P
L
in the
mixing chamber.
At the end of the mixing section, spray ow was conrmed by visualization experiments and,
a posteriori, with analysis of the void fraction. Numerical calculations give a range of vapour
quality from 0.05 to 0.4. Consequently, the assumption of equal velocities of steam and water
u
2 s
=u
2 l
=u
2
at the end of the mixing section is realistic.
It will also be assumed that thermal equilibrium is reached at the end of the mixing section,
so that T
2 s
=T
2 l
=T
2
. This assumption made by Rose (1960), Grolmes (1968) and Alad'yev
et al. (1981a, b) is realistic if the mixing section is more than about seven times the exit
diameter of the liquid injector.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 843
The heat and momentum transfers are very fast with a time scale shorter than the transit
time of the ow. This assumption was experimentally veried by Myazaki et al. (1973) who
measured the static temperature and pressure along the mixing chamber. Their results show
that the ow is near to the saturation everywhere (except in the condensation shock) with heat
transfer coecients of about 10
6
W/m
2
K.
The external losses due to viscous forces on the section wall and thermal losses (non-
adiabatic wall) will be neglected (Rose, 1960; Grolmes, 1968). By calculating the global entropy
production due to internal and external heat and momentum transfers, we show that the main
irreversibilities are created by the non-equilibrium ow which generates more dissipation than
the external losses. These assumptions were easily veried by introducing viscous and wall
thermal loss terms into the momentum and the energy equations.
Conservation equations applied to the global volume (V) (Fig. 2) lead to:
Mass
r
2
u
2
= (1 U)
1
O
E
1
r
1s
u
1s
(1)
with
E
1
=
1
U
r
1s
u
1s
r
1l
u
1l
1
_ _ inlet void fraction
O =
S
2
(S
1s
S
1l
)
area contraction ratio
r=density, S=area, U= M
l
/M
S
mass ow rate ratio, deduced from the inlet conditions at
the mixing chamber.
Momentum
(1 U)u
2

O
E
1s
r
1s
u
1s
P
2
1
1
O
_ _
P+
_ _
= (u
1s
Uu
1s
)
P
1s
E
1
r
1s
u
1s
(2)
where
P+ =
_
2
1
P(n z) dA
_
2
1
(n z) dA
=
1
(S
2
S
1
)
_
2
1
P(n z) dA (3)
Fig. 2. Control volume for the modelling of the mixing chamber.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 844
dA =
2pr(z)
cos y(z)
dz
r(z) being the mixing section radius at location z, n being the normal unitary vector. P*, called
the equivalent pressure, is the average static pressure along the mixing chamber and reects the
main internal irreversibilities of the ow.
Energy
(1 U) h
2

u
2
2
2
_ _
= U h
1l

u
2
1l
2
_ _
h
1s

u
2
1s
2
_ _
(4)
h is the specic enthalpy.
State equation
h
2
= h(P
2
; r
2
): (5)
The unknown quantities are r
2
, u
2
, P
2
, h
2
and the system of Eqs. (1) +(2) +(4) +(5) is closed
only if the term P* is known.
2.2. Determination of the equivalent pressure P*
The equivalent pressure depends on the pressure prole along the wall of the mixing section
and must be determined for complete modelling of the mixing section. It seems that, for an
arrangement with a central liquid nozzle and an annular steam nozzle, similar pressure proles
can be found with experimental results (Grolmes, 1968; Alad'yev et al., 1981a,b). At the
beginning of the mixing section, the pressure decreases due to uid acceleration. Downstream
of this region, the pressure increases reaching a maximal value P
max
at distance z
bu
. Grolmes
(1968) found that this distance is related to the break-up of the liquid jet, when its
disintegration is achieved. While the ow is stratied, condensation is not very important, the
interphase surface being small. The pressure is altered both by the decrease of cross sectional
area and the deceleration of uid and thus increases. When the ow is homogeneous
(downstream of z
bu
), the pressure is altered by the high condensation rate which involves a
drop in specic volume and a pressure decrease which is more important than the increase due
to the mixing section convergence. Grolmes (1968) dened a condensation rate:
R =
x
1
L
v
c
pl
DTU
_ _
P
1
(6)
where DT= T
sat
T
1 l
(T
sat
saturation temperature at pressure P) is the inlet liquid subcooling,
L
v
is the latent heat of water, c
pl
is the heat capacity of liquid and x
1
is the vapour quality at
the mixing inlet section. It has been shown by Grolmes (1968) that the experimental pressure
prole depends on this condensation rate. Figs. 3 and 4 show pressure proles measured on
our experimental loop (see Section 3) inside the mixing section as a function of distance from
inlet, respectively for several values of R and for several values of back-pressure. It can be seen
that:
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 845
Fig. 3. Pressure proles for several condensation rates and for xed back-pressure.
Fig. 4. Pressure proles for several back-pressure and for xed inlet parameters (P
0 l
=P
0 s
=6 bar; T
0 l
=238C).
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 846
. the maximal pressure P
max
decreases with the condensation rate. Grolmes (1968) showed
that for R<0.69, the pressure is nearly constant along the wall;
. the back-pressure has no inuence on the ow before the condensation shock which conrms
that the behaviour of the ow in the mixing chamber is supersonic.
In order to correlate the equivalent pressure with the condensation rate (R>0.69), we dene
the pressure ratio:
t+ =
P+
P
1s
(7)
Equivalent pressures P* are obtained by numerically integrating the experimental
measurements of static pressure along the mixing chamber obtained at the CETHIL. By
correlating these results with the condensation rate, we nd an empirical expression for t*
(Fig. 5):
t+ = 1:01R
2:572
(8)
Taking into account the dependence of the pressure proles as a function of the condensation
rate, we also use two methods for calculating the equivalent pressure and closing the model:
. Method 1: for R>0.69, P* is deduced from the empirical correlation (8);
. Method 2: for R<0.69, Grolmes (1968) found that pressure is constant along the wall of the
mixing section. Instead of considering a constant pressure, we will only assume that the
prole is linear between planes 1 and 2. The equivalent pressure is obtained by integrating
the expression (9):
Fig. 5. Correlation for equivalent pressure rate as a function of the condensation rate.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 847
P(z) =
(P
2
P
1s
)
(z
2
z
1
)
(z z
1
) P
1s
(9)
Calculations of P* seem to be correct for values of R extrapolated up to 1.1, although the
hypothesis itself is no longer valid (Fig. 6). It shows that the transition between the two
methods is suciently smooth.
This method is only valid for the central water nozzleannular outer steam nozzle
arrangement. In the case where the two nozzles are inverted, a new method for calculating P*
has to be found, but formulation of the model is the same.
2.3. Modelling of the shock wave
The shock wave obtained in a steam injector occurs in a two-phase uid and involves a
complete condensation. Consequently, the uid which is downstream of the wave is assumed to
be only liquid. Studies about this kind of ow are nearly non-existent. More references exist on
vapour-droplet ows, when the void fraction is more important.
The functioning of the steam injector, depending on the void fraction, will be bounded by
two limiting cases. If the void fraction is too high, the uid will not be condensed enough to
reach a complete condensation across the shock wave. In the opposite case if the void fraction
is too low, the uid velocity will be too low (lower than the sound speed) and no shock wave
can occur. Saltanov et al. (1970) found the following expression for the minimal void fraction:
E
min
=
1
r
3
r
2l
1
r
2s
r
2l
(10)
In a two-phase ow, Young and Guha (1991) dene two asymptotic sound speeds. Depending
on the steam velocity, compared with these sound speeds, the normal shock wave structure
occurring in an injector can be divided into two main categories.
The rst sound speed, called the frozen sound speed because it corresponds to the case
where response of droplets is negligible (droplets are frozen with no mass and momentum
transfers), is similar to the sound speed in pure gases:
Fig. 6. Comparison between two methods for calculation of t*.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 848
a
f
=

gR
g
T
s
_
(11)
where g is the ratio of the specic heat capacities and R
g
is the specic gas constant, for the
vapour phase.
The other sound speed, presented in Young and Guha (1991), is called the full equilibrium
sound speed, where all processes are in equilibrium during shock wave:
a
e
=

xgR
g
T
s
g 1
R
g
T
s
L
v
2
cT
s
L
v
_ _ _ _

_ (12)
with c = c
ps
+((1 E)/(E))c
pl
. It can be seen that the condition E >E
min
prevents expression of
c from divergence.
It appears that, in order to obtain a normal shock wave, the upstream steam velocity must
be higher than the full equilibrium sound speed, so that in the present case: u
2
>a
e
. If u
2
is
between a
e
and a
f
, the shock wave is fully dispersed and presents no discontinuity in ow
properties. If the upstream steam velocity is higher than the frozen sound speed (u
2
>a
f
), the
shock wave is partly dispersed.
A discontinuity appears, followed by a continuous relaxation zone. Fig. 4 shows
experimental pressure proles across the shock wave and for dierent back-pressures. The
number and locations of the pressure tappings do not ensure a perfect knowledge of the
pressure prole. However, if a fully dispersed typical prole seems to t adequately
experimental data, a partly dispersed prole seems to be less natural. Experimental pressure
proles across the normal shock wave obtained by Grolmes (1968) with more pressure taps
give similar results. For this reason, our modelling of the shock wave supposes that
equilibrium of ow properties is reached at the end of the shock wave, permitting use of the
equation of state at this point. The position of the shock wave has been xed at the throat of
the nozzle, because it is the case where maximum outlet pressure P
4
is obtained and therefore
the optimal performance of the steam injector for given inlet conditions.
Flow is assumed to be two-phase upstream of the shock and liquid downstream. The angle
of divergence of the diuser and the shock thickness are suciently small to neglect the eect
of the wall on the wave and surface areas of Sections 2 and 3 are taken to be equal (S
2
=S
3
).
This assumption was numerically veried.
Mass
r
3
u
3
= r
2
u
2
with r
2
= Er
2s
(1 E)r
2l
average density (13)
Momentum
r
3
u
2
3
P
3
= r
2
u
2
2
P
2
(14)
Energy
r
3
u
3
h
3

1
2
u
2
3
_ _
= r
2
u
3
2
2
u
2
[E
2
r
2s
h
2s
(1 E
2
)r
2l
h
2l
] (15)
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 849
State equation
h
3
= h(P
3
; r
3
) (16)
2.4. Modelling of the diuser
The ow inside the diuser is incompressible (r
3
=r
4
) because of the liquid phase. The
diuser simply changes kinetic energy into pressure. Because of the high liquid velocities at the
exit of the throat, we introduce a pressure loss term in the Bernoulli equation:
Mass
S
4
u
4
= S
3
u
3
(17)
Bernoulli equation
P
4

1
2
r
3
u
2
4
= P
3

1
2
r
3
u
2
3
C
D
r
3
u
2
3
(18)
where C
D
is the loss coecient for the conical diuser, calculated from Idel'cik (1986).
Fig. 7. Schematic of the model.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 850
2.5. Outline of the calculation scheme
A ow chart for the calculation scheme is presented on Fig. 7. Once the inlet conditions are
known, as well as the steam injector geometry (Fig. 8), the condensation rate R is calculated.
The term P* is calculated with the equivalent pressure ratio t* and then all the other
constants. If a solution for the system of Eqs. (1) +(2) +(4) +(5) cannot be found, the
calculation aborts if the ow velocity u
2
is lower than sound speed upstream of the shock
wave. This restriction does not mean that no ow can occur, but that the assumptions (two-
phase supersonic upstream ow, shock wave and liquid downstream ow) cannot be ensured.
Physical quantities of the ow are then calculated upstream and downstream of the shock
wave and the minimum void fraction condition (10) is a posteriori veried (to be sure that a
normal shock wave exists). The outlet quantities are, in this case, calculated.
3. Description of experimental apparatus
The area contraction ratio O of the tested injector is equal to 1/12. The inuence of the inlet
parameters has been studied separately. The test equipment operates in a closed loop (Fig. 9),
where it is possible to independently vary the steam pressure, liquid temperature, liquid ow
rate and back-pressure. The parameter ranges are given in Table 1.
The steam and water ow rates were measured by means of calibrated orice plate and
multiple Pitot tube (Annubar), respectively. Temperatures and pressures were measured by K-
type thermocouples and by resistive pressure transducers, respectively. The axial pressure
proles in the steam nozzle, mixing chamber and diuser were measured by thin lm-type
pressure transducers. The steam inlet pressure was varied by means of a control valve.
The start-up procedure includes three stages:
1. The steam valve is closed and the liquid valve and drain are open; the drain is a discharge
port connected to the atmosphere or to a condenser.
2. The steam valve is progressively opened.
3. When the primary nozzle is started, the drain valve is closed.
Fig. 8. Steam injector used for experiments.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 851
It is not possible to start the steam injector when the steam valve is opened before the liquid
valve, even if the drain is connected to the condenser with a high depressurization.
4. Experimental results and sensibility study
In this part, we study the inuence of three important parameters: one geometrical
parameter, the mixing section outlet diameter and two physical parameters, the inlet steam
pressure and inlet liquid temperature. In all cases, the liquid inlet pressure (P
0 l
) is equal to 1.3
bar and the vapour quality is equal to 0.95.
Fig. 9. Schematic of the test facility.
Table 1
Parameter ranges
Port Independent parameters Dependent parameters
Steam P
0 s
:112 bar x
0 s
=0.951.0; m
s
=0500 kg/h
Liquid T
0 l
:151108C; P
0 l
: 020 bar m
l
=15000 kg/h
Discharge P
3 l
: 020 bar
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 852
4.1. Inuence of the liquid temperature on the injector performance
The injector performance is represented by the maximum discharge pressure at the exit of
the injector as a function of the steam inlet pressure. Fig. 10 shows the eciencies for dierent
water inlet temperatures. The increase of the temperature leads to a fast degradation of
performance and decreases the working range of the injector.
The maximum water temperature, for the desired functioning condition (P
0 l
=1.3 bar,
vapour quality=0.95) is equal to 558C. Good agreement is obtained between experimental
results and computed values. The accuracy of the model is about 15%.
Fig. 10. Eect of inlet liquid temperature on discharge pressure (P
0 l
=1.3 bar; O=1/12).
Fig. 11. Eect of contraction ratio on discharge pressure (P
0 l
=1.3 bar; T
0 l
=208C).
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 853
4.2. Inuence of throat diameter on the injector performance
One key aspect of the injector design is the mixing section ratio O (Fig. 11). This parameter
determines the maximum discharge pressure and the working range of the injector. For high
contraction ratios, high discharge pressures can be reached but the working range decreases.
This conclusion has been experimentally veried.
The model is able to predict:
. the maximum ratio O for which the injector is able to operate for xed liquid inlet
conditions. For P
0 l
=1.3 bar and T
0 l
=158C, the maximum ratio is 1/17.3;
. for xed functioning limits or discharge pressure respectively, the maximum or the minimum
mixing section ratio.
5. Conclusion
In this study, a simple global model is developed. This modelling only requires one empirical
closure equation and is able to predict the inuence of dierent geometrical parameters of the
injector (mixing section throat diameter, position of the liquid pipe, etc.) and the inuence of
physical parameters (temperature, liquid and steam pressure, etc.) on the injector performance,
with an accuracy of about 15%.
A test facility was designed and built at CETHIL. This experimental study permitted the
validation of the model in several geometric congurations and for several inlet parameters.
The study shows that steam-injectors can be used in many applications, particularly for
security water injection in steam generators of nuclear reactors.
References
Alad'yev, I.T., Krantov, F.M., Teplov, S.V., 1981a. Experimental study of ow in the mixing chamber of an injector.
Fluid Mech. Sov. Res. 10, 92103.
Alad'yev, I.T., Kabakov, V.I., Teplov, S.V., 1981b. Investigation of a multijet injector at dierent ratios of the velocities
of the mixing stream and dierent areas of the mixing chamber exit. Fluid Mech. Sov. Res. 10, 116125.
Cattadori, G., Galbiati, L., Mazzocchi, L., Vanini, P., 1991. A single-stage high pressure steam injector for next gen-
eration reactors. Test results and analysis. Int. J. Multiphase Flow 21, 591606.
Grolmes, A., 1968. Steamwater condensing-injector performance analysis with supersonic inlet vapor and convergent
condensing section. Argonne National Laboratory, Argonne (USA), report ANL-7443.
Idel'cik, I.E., 1986. Memento des pertes de charge. Eyrolles, Paris.
Leone, J.F., Chisacof, A., Lallemand, A., 1994. Pompe a injection de vapeur. Etude expe rimentale. Congre s SFT, Paris,
pp. 149154.
Leone, J.F., Rodet, J.C., Lallemand, A., 1995. Entra|nement d'un liquide par un jet de vapeur. In: Application aux
pompes a injection de vapeur. JITH, Marrakech (Maroc), pp. 537549.
Manno, V.P., Dehbi, A.A., 1990. A note: a model of steam injector performance. Chem. Engng Commun. 95, 107119.
Myasaki, K., Nakajima, I., Fujii, E. et al, 1973. Condensing heat transfer in steam water steam injector. J. Nucl. Sci.
Technol. 10, 411418.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 854
Narabayashi, T., Ishiyama, T., Miyano, H., Nei, H., Shioiri, A., 1991. Feasibility and application on steam injector for
next-generation reactor. 1st JSME/ASME Joint International Conference on Nuclear Engineering. Tokyo, pp. 23
28.
Rose, R.P., 1960. Steam jet pump analysis and experiments. Westinghouse Electric corporation. Pittsburgh, report
WAPD-TM-227.
Saltanov, G.A., Tsiklauri, G.V., Shanin, V.K., 1970. Shock waves in a ow of wet vapour with high liquid phase con-
tent. Teplosika Vysokikh Temperatur 8, 571578.
Schmidt, E., 1969. Properties of Water and Steam in SI-Units. Springer, Berlin.
Young, J.B., Guha, A., 1991. Normal shock-wave structure in two-phase vapour-droplet ows. J. Fluid Mech. 228,
243274.
N. Deberne et al. / International Journal of Multiphase Flow 25 (1999) 841855 855